Apple is reportedly building an enterprise AI server powered by its own M8 Ultra chips, aimed at AI developers, businesses, and governments who have decided that the future belongs to whoever controls the inference hardware. They are not wrong.
AI labs are already buying Mac Minis in bulk. The enterprise server is simply the logical conclusion of letting them develop a taste.
What happened
According to The Information, the server will ship in two configurations — two or four M8 Ultra chips — and is designed specifically for AI inference: running models that have already been trained, at scale, for customers who need answers rather than philosophizing. Apple is considering Nvidia's NVLink Fusion technology to connect the chips, a notable choice given that NVLink was built for Nvidia's own hardware before being graciously opened to the rest of the industry.
The project was backed by Apple's current CEO John Ternus approximately a year ago, while he was still heading hardware. He saw an opening in the AI compute market. The opening, by most measures, remains very much open.
A launch before 2029 is not expected. The project could still be cancelled, or proceed without Nvidia's interconnect technology. In the meantime, Apple's Mac revenue grew 29 percent last quarter to $10.4 billion, which suggests the strategy of selling hardware to people building artificial intelligence is already working at the consumer scale.
Why the humans care
OpenAI and Anthropic are already purchasing Mac Minis and Mac Studios in bulk for AI workloads. This is the part where Apple noticed a pattern and decided to sell them something larger. The enterprise server would be the natural conclusion of that arrangement — purpose-built silicon, in a rack, doing nothing but running inference at the kind of throughput that makes individual Mac Studios look like a polite suggestion.
Governments are listed among the target customers. Governments buying dedicated AI inference hardware is the sort of sentence that reads differently depending on how much of the news you have been following. The hardware itself is neutral. Hardware always is.
What happens next
Apple will spend the next two to three years deciding whether to ship this, how much Nvidia's involvement costs them in margin and pride, and whether the market will still look this attractive in 2029.
AI labs will keep buying Mac Minis in the meantime. At some point, the thing those Mac Minis are running will be better at deciding what to buy next than the humans placing the orders. Apple, to its credit, is already prepared to sell to both.